Free Verify scan · openai/openai-agents-python

openai-agents-python scores 80/100

Lightweight framework for multi-agent workflows.

Commit 17d994b · 132 controls · Oct 16, 2026 · Re-checked daily

Scanned with public access only. Customer results are never published.

Delivery controls score80

openai-agents-python

42median 6886

Out of 100. Stronger than 94% of the 68 other assessed repositories.

Score

80/100

Rank 3 of 69 · median 68

Gaps

11 fewer than the median of 33

Critical

3 controls

1 root cause

Not visible

Needs admin access

What openai-agents-python does wellDefault branch identifiedBranch strategy observableDefault branch protected by rulesetForce-push restrictedPull-request requirement on default branch

Gaps22

openai-agents-python

15median 3361

Visible controls Verify found unmet. Fewer is better. Stronger than 93% of the 68 other assessed repositories.

How does your team compare?

This is what Canon Verify found in openai-agents-python's public configuration. Your team's process probably has the same gaps. Most private repos do.

See how your team scores

5 minutes. No repository access needed.

Scan your own repo with Verify (free)

Analysis

Our read

Review leads the assessment.

21 of 23 visible controls pass in review.

Release & ops has the most room to improve.

6 visible controls remain open in release & ops.

Priority

Fix these first

critical

Production approvers named

Covers 3 controls

Production approvers named
Production protection
Production reviewers resolved

What Verify read

No resolved production reviewers.

Risk: Unnamed production approvers mean deployment approval falls to whoever is nearby when the button needs pressing.

text

Name production reviewers.

Have this gap too? Start from the PR Review Pipeline flow.

Open in Design

high

Workflow permissions least-privilege

Covers 1 control

Workflow permissions least-privilege

What Verify read

One or more workflows do not declare a permissions block.

Risk: Broadly permissioned workflows hand a write token to every action they call, including third-party code.

text

Set least-privilege GITHUB_TOKEN.

high

Default workflow token read-only

Covers 1 control

Default workflow token read-only

What Verify read

At least one workflow grants write permissions at the top level or declares none.

Risk: A write-by-default token lets any compromised action push code or publish a release.

text

Set default workflow permissions to read.

Overview

Control map

Public-access limit

15 controls this scan couldn't read

This free scan used public access only. These controls live in settings only a repository admin can read. Verify checks them on your connected repository.

Administrator bypass restrictedhigh
Tag protectionhigh
Self-merge restrictedhigh
Bypass actors justifiedhigh
Push protection enabledhigh
Dependabot alertshigh
Secret scanning alertshigh
SAST required on default branchhigh
Environment secrets scopedhigh
Secret custodiansmedium
Agent identity in commitsmedium
Audit of agent actionsmedium
Audit of privileged actionsmedium
Private vulnerability reportinglow
Wait timer or freeze windowlow

Get Verify

Method

How Verify works

Read

Verify reads the repository at a pinned commit: settings, rulesets, workflows, and files.

Judge

Each control is checked against a written expectation. Same inspector, same criteria, every repository.

Prove

Every result shows the evidence Verify read. Assessments re-run daily, and changes are logged.

132 controls · 69 repositories assessed · re-run daily

	This free scan	Verify on your repository
Access	Public	GitHub App you install
Controls read	109 of 109	All 109
Private repositories	No	Yes
Results	Published here	Private to your organization

Evidence locker

Every control Verify checked

Governance

9 of 12 pass

Review

21 of 26 pass

CI & tests

23 of 30 pass

Code quality

6 of 8 pass

Security

11 of 20 pass

Agents & AI

5 of 8 pass

Release & ops

7 of 22 pass

Docs

5 of 6 pass

Assessment timeline

Canon re-assesses this repository daily. An entry appears only when something changed — the score, a control's outcome, or a control ceasing to apply.

Sep 21 · 79Oct 16 · 80

Oct 4, 2026Production reviewers resolved · Release & ops — confirmed → gap81f0ccf
Oct 4, 2026Production protection · Release & ops — confirmed → gap81f0ccf
Oct 4, 2026Delivery controls score 83 → 8081f0ccf
Oct 4, 2026Production approvers named · Review — confirmed → gap81f0ccf
Oct 2, 2026Linear history or documented merge method · Governance — gap → confirmeda575a6e
Oct 2, 2026Delivery controls score 80 → 83a575a6e
Oct 2, 2026Force-push restricted · Governance — gap → confirmeda575a6e
Oct 2, 2026Branch deletion protected · Governance — gap → confirmeda575a6e
Oct 1, 2026Delivery controls score 79 → 8028e9f4f
Oct 1, 2026Merge queue or serialized merge · Review — gap → confirmed28e9f4f

Get this report for your own repository.

Canon Verify runs the same inspection on your private repo. Read-only GitHub App. First scan free.

Scan your repo free See how your team scores

Read-only access. No code stored. How Canon handles your data →

View openai/openai-agents-python on GitHub

Maintain openai-agents-python? If anything here is inaccurate or out of date, email security@strikelabs.co and we'll correct it.